Customer Data Integration: Reaching a Single Version of the Truth

A CDI TAXONOMY

Not all CDI is created equal. Whether companies write their own CDI programs or purchase CDI solutions from vendors, the way CDI processes data can vary. Although there are variations and interpretations of the different types of CDI, there are three acknowledged types: registry, persistent, and hybrid. We’ll refer to them elsewhere in the book as we further explore issues around CDI functionality and architecture, but we’ll introduce them briefly here.

Registry Style

With a registry style, the customer data remains on the source system and a series of pointers or data linkage details are stored physically on the CDI hub. Registry CDI acknowledges that the data “lives” on the source system (s), simply accessing and reconciling it for the purpose of creating the authoritative view or record of the customer.

The registry style (also known as the “reference” style) is arguably the most straightforward type of CDI to implement. The data remains on the source system and the CDI hub retains reference details, or pointers, to the data on the source. This minimizes the involvement of source system owners and programmers who might already be nervous about outside access of their data.

As Exhibit 2.4 shows, the CDI hub recognizes the unique customer record in each source system.

The hub assigns a unique identifier to that customer, Customer Number 1, and records the unique identifier of that record in the source system. The hub maintains a list of data elements and ...